#Google DeepMind

'Nano Banana Pro, built on Gemini 3 Pro, delivers text-accurate, multilingual and production-ready image generation with studio controls and upscaling to 4k. It grounds visuals in reasoning and Search to turn data and documents into information-rich imagery.'

'Google teams developed g-AMIE a guardrailed multi agent diagnostic system that separates intake from clinician authorization achieving higher documentation quality and faster oversight in simulated clinical exams'

Google DeepMind introduces Genie 3, a groundbreaking AI capable of creating diverse, interactive virtual worlds from simple text descriptions, enabling new possibilities in gaming, robotics, and XR.

Google DeepMind's AlphaEarth Foundations introduces a revolutionary AI-based 'virtual satellite' that integrates diverse Earth observation data into high-resolution, up-to-date maps, overcoming traditional data scarcity challenges.

Google DeepMind introduces Aeneas, an AI-powered tool that enhances restoration, dating, and contextual understanding of ancient Latin inscriptions, significantly aiding historians and researchers.

Google DeepMind and collaborators introduce Crome, a causal framework that improves reward modeling robustness in LLM alignment by using counterfactual data augmentation to tackle reward hacking issues.

Google DeepMind's AlphaGenome is a novel deep learning model that predicts the regulatory impact of DNA mutations across multiple biological modalities with high precision, outperforming existing models in genomic tasks.

'Google DeepMind and partners have developed Motion Prompting, a novel method allowing detailed control of video generation via motion trajectories, showcased at CVPR 2025.'

Google DeepMind introduces Gemma 3n, a compact multimodal AI model optimized for real-time on-device performance, delivering faster response times and enhanced privacy on mobile platforms.

Google DeepMind researchers developed a reinforcement learning fine-tuning method that significantly improves large language models' ability to act on their reasoning, reducing the gap between knowledge and action.

Google DeepMind introduces QuestBench, a benchmark designed to evaluate how well large language models identify missing information in complex reasoning tasks and generate necessary clarifying questions.

Google DeepMind CEO Demis Hassabis warns that AGI could emerge by 2035, urging global cooperation to prepare society for its profound impact and risks.